through the TensorRT engine Construct method, some complex layer fusions can't be quickly identified. TensorRT-LLM optimizes these making use of plugins which can be explicitly inserted to the community graph https://allytech.net.in/courses/python-training-in-bangalore/